HAL Id: hal-01788483
https://hal.archives-ouvertes.fr/hal-01788483
Submitted on 23 May 2018
HAL is a multi-disciplinary open access archive for the deposit and dissemination of sci- entific research documents, whether they are pub- lished or not. The documents may come from teaching and research institutions in France or abroad, or from public or private research centers.
L’archive ouverte pluridisciplinaire HAL, est destinée au dépôt et à la diffusion de documents scientifiques de niveau recherche, publiés ou non, émanant des établissements d’enseignement et de recherche français ou étrangers, des laboratoires publics ou privés.
Flow Cytometric Single-Cell Analysis for Quantitative in Vivo Detection of Protein–Protein Interactions via Relative Reporter Protein Expression Measurement
Lina Wu, Xu Wang, Tian Luan, Jianqiang Zhang, Emmanuelle Bouveret, Xiaomei Yan
To cite this version:
Lina Wu, Xu Wang, Tian Luan, Jianqiang Zhang, Emmanuelle Bouveret, et al.. Flow Cytometric Single-Cell Analysis for Quantitative in Vivo Detection of Protein–Protein Interactions via Relative Reporter Protein Expression Measurement. Analytical Chemistry, American Chemical Society, 2017, 89 (5), pp.2782-2789. �10.1021/acs.analchem.6b03603�. �hal-01788483�
Flow Cytometric Single-Cell Analysis for Quantitative in
1
Vivo Detection of Protein–Protein Interactions via Relative
2
Reporter Protein Expression Measurement.
3
Lina Wua,1, Xu Wanga,1, Tian Luana, Jianqiang Zhanga, Emmanuelle Bouveretb and
4
Xiaomei Yana,2
5
aThe MOE Key Laboratory of Spectrochemical Analysis & Instrumentation, The Key
6
Laboratory for Chemical Biology of Fujian Province, Department of Chemical
7
Biology, College of Chemistry and Chemical Engineering, Xiamen University,
8
Xiamen, Fujian 361005, P. R. China
9
bLaboratory of Macromolecular System Engineering (LISM), Institute of
10
Microbiology of the Mediterranean (IMM), Aix-Marseille Univ and Centre National
11
de la Recherche Scientifique (CNRS), Marseille 13402, France
12 13
1L. W. and X. W. contributed equally to this work.
14
2To whom correspondence should be addressed. E-mail: xmyan@xmu.edu.cn.
15 16
Classification:
17
Major: Biological Sciences
18 19
2 / 26
Abstract:
20
Cell-based two-hybrid assays have been key players in identifying pairwise
21
interactions, yet quantitative measurement of protein-protein interactions in vivo
22
remains challenging. Here, we show that using relative reporter protein expression
23
(RRPE), defined as the level of reporter expression normalized to that of the
24
interacting protein measured in single cell, quantitative analysis of protein interactions
25
in bacterial adenylate cyclase two-hybrid (BACTH) system can be achieved. A
26
multicolor flow cytometer was used to measure simultanously the expression levels of
27
one of the two putative interacting proteins and the β-galactosidase (β-gal) reporter
28
protein upon dual immunofluorescence staining. Single cell analysis revealed that for
29
every bacterial culture co-transformed with the two-hybrid plasmids, there exist two
30
cell populations with or without the expression of interacting protein and reporter
31
protein. Using Pal and TolB protein as the model of an interacting bait and pray pair,
32
the RRPE was found to be constant regardless of the inoculation colonies and the
33
cultivation time. Decreased RRPE was detected for TolB mutants with two
34
N-terminal truncations (TolB Δ22-25 or TolB Δ22-33), suggesting that the RRPE was an
35
intrinsic characteristic associated with the binding strength between the two
36
interacting proteins. This hypothesis was verified with acid base coiled coils formed
37
by two α-helices of various binding affinities, for which the measured RRPE
38
progressively decreased as the affinity decreased. Several useful applications of our
39
RRPE-BATCH method can be expected for the quantitative detection, strength
40
comparison, and affinity ranking of pairs of interacting proteins.
41
Keywords: protein-protein interaction, yeast two-hybrid system, bacterial two-hybrid
42
system, flow cytometry, binding affinity
43 44
Significance Statement
45
Assessing the intrinsic affinities for interacting proteins is of fundamental importance
46
to explore and understand protein-protein interactions. Using a bacterial two-hybrid
47
system (the BACTH system), we developed a quantitative method for the detection
48
and affinity ranking of protein-protein interactions. By measuring the expression level
49
of both the reporter protein and interacting protein at the single-cell level via
50
immunofluorescent staining and flow cytometry, we found that the relative reporter
51
protein expression (RRPE) is characteristic of the interacting protein pair and
52
correlates with their binding affinity. This method can provide an efficient tool in
53
prioritizing a large number of putative interacting proteins for following analyses.
54 55
4 / 26
Protein-protein interactions are involved in virtually every cellular process, and their
56
study is crucial in revealing protein functions, deciphering protein interaction
57
networks, and identifying novel therapeutic targets (1, 2). Among numerous
58
methodologies developed for protein interaction study, the yeast two-hybrid (Y2H)
59
system is the most commonly used binary method for measuring direct physical
60
interactions between two proteins, and has been estimated to account for over 50% of
61
protein-protein interactions described in PubMed (3-5). This powerful in vivo
62
approach interrogates two proteins, called bait and prey, one fused to a DNA-binding
63
domain and the other fused to a transcriptional activation domain of a transcription
64
factor and expressed in yeast. If the two proteins interact in the system, they
65
reconstitute a functional transcription factor that induces the transcription of a reporter
66
gene, whose output can be measured as growth of yeast colonies on selective medium
67
or as blue coloration in a β-galactosidase (β-gal) assay. Although the Y2H system has
68
made significant contribution to the discovery of protein-protein interactions and the
69
interactome networks (6-8), both the false positive and false negative rates are
70
relatively high, and all the interactions are forced to occur in the yeast nucleus and are
71
thus not suitable for protein interaction involving membrane proteins and cytosolic
72
proteins (7, 9).
73
To overcome the limitations of Y2H system, a bacterial equivalent of the
74
two-hybrid system was developed based on functional complementation of the
75
catalytic domain of Bordetella pertussis adenylate cyclase (BACTH) (10, 11). This
76
leads to cAMP synthesis, which in turn, triggers the expression of several resident
77
genes such as the lactose or maltose operons. Particularly, this technique enabled the
78
study of membrane proteins because cAMP is a diffusible molecule and the BACTH
79
system does not require the hybrid proteins to be located in the nucleus as that of Y2H
80
(12). However, the BACTH system as well as the Y2H system is not suitable for the
81
quantitative measurement of pairwise protein interactions due to the lack of
82
understanding of how the strength of the interactions correlate with the level of
83
reconstituted reporters (13).
84
Flow cytometry is a well-established tool for the rapid, quantitative, and
85
multiparameter analysis of single cells. Employing a codon-optimized yeast enhanced
86
green fluorescent protein (yEGFP) as the reporter, Chen et al. developed a high
87
throughput approach to study protein-protein interactions inside the cell via flow
88
cytometric measurement (14). Through the development of a yeast surface two-hybrid
89
(YS2H) system, Hu et al. reported quantitative flow cytometric measurement of
90
protein-protein interactions via the secretory pathway (13). On the other hand,
91
anchored periplasmic expression (APEx) bacterial two-hybrid system has been
92
developed for the study of protein pairs on the basis of affinity or expression (15, 16).
93
In both the YS2H and the APEx two-hybrid systems, the bait protein has to be
94
produced at the surface of the cell, and the tag-fused prey protein has to be secreted in
95
solution. Then, the strength of bait-pray interaction can be measured via antibody
96
binding to the epitope tag appended to the prey protein. Compared to the surface
97
bacterial two-hybrid system, the classic BACTH system is a well-established and
98
much simpler approach for protein-protein interaction studies (10, 11). Particularly,
99
6 / 26
the signaling cascade in the BACTH system ensures higher sensitivity for the weak
100
and transient interactions. Therefore, a quantitative approach for the BACTH system
101
shall greatly advance the protein-protein interaction study due to the general
102
applicability of the BACTH system and its sensitivity for low affinity interaction.
103
Herein, we demonstrate that through flow cytometric detection and
104
immunofluorescent staining, the relative reporter protein expression (RRPE), defined
105
as the normalized report protein expression to that of the interacting protein in a
106
single cell, can be used to quantitatively estimate the binding strength between two
107
interacting proteins in the classical BACTH system. This feature allowed us to
108
confirm interacting pairs of proteins, investigate determinant residues in
109
protein-protein interaction, and compare interaction strength of different pairs. The
110
RRPE-BACTH method described here provides a practical and powerful method for
111
the rapid and quantitative in vivo measurement of protein-protein interactions.
112
Results and Discussion
113
Design of the BACTH System for Flow Cytometric Analysis. Scheme 1 illustrates
114
the experimental design of the flow cytometric BACTH system. Two compatible
115
plasmids carrying the hybrids with T25 and T18 domains respectively were
116
co-transformed into the reporter strain cya- E. coli BTH101. Between T25 or T18
117
domain and the hybrid proteins, His and Flag tags were inserted respectively, and
118
were used to follow the expression of the hybrid proteins. The interaction of the
119
hybrid proteins results in a functional complementation between T25 and T18
120
fragments, which reconstitutes the activity of adenylate cyclase and leads to cAMP
121
synthesis. The produced cAMP interacts with the catabolite activator protein (CAP)
122
and the cAMP/CAP complex binds to the promoter and regulates the transcription of
123
lacZ gene coding for the β-galactosidase (β-gal) reporter expression. β-gal was
124
specifically labeled green with rabbit-anti-β-gal antibodies and FITC-conjugated goat
125
anti rabbit IgG. Meanwhile, anti-His/Flag mouse monoclonal antibody and DyLight
126
649-conjugated goat anti mouse IgG were used to label the hybrid proteins red. Upon
127
dual immunofluorescence labeling, the bacterial sample was analyzed by flow
128
cytometer. FITC and DyLight 649 fluorophores can be excited by the 488 nm and 640
129
nm lasers respectively, and the emitted green and red fluorescence signals were
130
detected concurrently on the FL1 and FL2 fluorescence channels. Therefore, for a
131
bacterial two hybrid sample, the expression level of both the β-gal reporter and the
132
hybrid proteins can be detected and quantified simultaneously.
133
134
8 / 26
Scheme 1. Depiction of protein-protein interaction study at the single-cell level based 135
on the BACTH system, dual immunofluorescent staining, and flow cytometric 136
analysis.
137
Flow Cytometric Detection of Protein-Protein Interaction by the BACTH System
138
via Immunofluorescent Staining of β-gal Reporter Protein. Pal and TolB, two
139
proteins involved in maintaining the integrity of bacterial outer membrane were
140
chosen as the protein-protein interaction model (17). In the BACTH system, both the
141
interacting proteins and reporter proteins are produced in the cytoplasm of bacteria,
142
and the antibodies need to traverse the bacterial cell wall and membrane for target
143
staining. Therefore, much efforts have been devoted to optimize the
144
immunofluorescence labeling procedures including fixation, permeabilization, and
145
staining (see Materials and Methods). After immunofluorescent staining of β-gal
146
reporter protein with FITC, the samples were analyzed on the flow cytometer. Fig. 1
147
shows the bivariate dot-plots of side scatter intensity versus FITC fluorescence
148
intensity obtained for E. coli BTH101 co-transformed with plasmids
149
pUT18C-linker/pKT25-linker (negative control, no interaction),
150
pUT18C-pal/pKT25-tolB (interacting proteins without additional tag), and
151
pUT18C-Flag-pal/pKT25-His-tolB (tagged interacting proteins), respectively. Two
152
distinct populations with different green fluorescence intensity were observed for
153
bacterial samples co-transformed with plasmids containing the interacting Pal/TolB
154
pair proteins regardless the presence of Flag/His tag or not (Fig. 1, B and C). A
155
discriminant line between these two populations was drawn on the FL1 channel for
156
easy discrimination, defining two regions P1 and P2. For the negative control sample,
157
approximately 94.8% of the cells fall in the P1 region (Fig. 1A), whereas for cells
158
co-transformed with plasmids containing the interacting proteins with or without tags,
159
about 34.8% and 42.4% of the cells fall in the P2 region. The fluorescence
160
distribution histograms of these three samples are plotted in Fig. 1D. Comparable
161
median fluorescence intensities (MFI) for cells falling in the P1 region were observed
162
for all three samples, suggesting that for bacterial culture co-transformed with two
163
plasmids carrying Pal and TolB genes, there exists a large fraction of cells (about 40%)
164
in which the β-gal reporter protein cannot be detected. Meanwhile, events residing in
165
the P2 region can be ascribed to cells that co-express Pal and TolB interacting
166
proteins inside a single cell which leads to the expression of β-gal reporter. Because
167
the interaction between Pal and TolB is robust and well characterized, the fraction of
168
the cells that are similar to the negative control may be ascribed to the lack of
169
expression of either one or both of the interacting proteins. As plasmid loss is
170
excluded by the use of antibiotics in the culture media, this all-or-nothing
171
phenomenon can only be explained by the bistability in the lactose utilization network
172
of E. coli. In the BACTH system, the reporter gene and hybrid plasmids are both
173
regulated by the wild type Plac promoter (18). For cells falling in the P2 region, the
174
MFI are 3135 and 2539 for bacterial cultures transformed with interacting protein
175
genes without and with Flag/His tag, respectively, suggesting that tag insertion to the
176
C-terminal of the two proteins did not prevent the interaction. The observation of two
177
populations with completely different behaviors regarding reporter β-gal expression
178
10 / 26
highlights the importance and need of single-cell analysis for the BACTH system. In
179
contrast to the ensemble-averaged measurement by spectrophotometers, flow
180
cytometric analysis can reveal the inherent heterogeneity of bacterial populations in
181
β-gal expression that will provide more insights into the BACTH system.
182
183
Fig. 1. Flow cytometric analysis of protein-protein interaction by the BACTH system 184
upon β-gal immunofluorescent staining. Bivariate dot-plots of side scatter intensity 185
(SSC) versus FITC fluorescence intensity (FL) for E. coli BTH101 co-transformed with 186
plasmids of A) pUT18C-linker/pKT25-linker, B) pUT18C-pal/pKT25-tolB, and C) 187
pUT18C-Flag-pal/ pKT25-His-tolB, respectively. D) FITC green fluorescence 188
distribution histograms of β-gal for E. coli BTH101 with no interaction (red, case A), 189
with expression of interacting proteins (blue, case B), and with expression of tagged 190
interacting (orange, case C).
191
Simultaneous Measurement of the Expression of Interacting Protein and
192
Reporter Protein. In order to study the relationship between the expression of β-gal
193
reporter protein and the expression of hybrid proteins, the Flag or His tag of one
194
interacting protein and the β-gal reporter protein were immunofluorescently labeled
195
with Dylight-649 and FITC, respectively. DyLight-649 was chosen to label the tag
196
fragment because it can be efficiently excited by the 640 nm laser, which avoids
197
spectral cross-talk with FITC. The green and red fluorescence signals were detected
198
on the FL1 and FL2 channel, respectively. Figs. 2A and 2B show the bivariate
199
dot-plots of β-gal reporter protein versus TolB expression (via His tag labeling) or Pal
200
expression (via Flag tag labeling), respectively. Quadrant gates were created for all
201
the samples. For the negative control with E. coli BTH101 co-transformed with
202
pUT18C-linker/pKT25-linker (no interacting protein expression), most cells (94.0%
203
and 93.9%) fall into the Q4 region (Figs. 2A1 and 2B1) with negligible fluorescence
204
on both the green and red fluorescence channels. For E. coli BTH101 co-transformed
205
with plasmids encoding interacting proteins but without tags
206
(pUT18C-pal/pKT25-tolB), 32.6% and 32.0% of the cells fall into the Q1 region due
207
to the expression of β-gal reporter protein. The two population phenomenon is similar
208
to the one observed in Fig. 1. For E. coli BTH101 co-transformed with plasmids
209
encoding interacting proteins with tags (pUT18C-Flag-pal/pKT25-His-tolB), 53.6%
210
and 54.4% of the cell population resides in the Q4 region (neither the expression of
211
β-gal nor the expression of interacting protein), while 42.5% and 41.2% of the cells
212
fall in the Q2 region indicating concurrent expression of the interacting proteins and
213
the β-gal reporter (Figs. 2A3 and 2B3). The similar ratios of Q2 reveal that only the
214
co-expression of the interacting proteins leads to the expression of the reporter gene.
215
Because expression of all of them are driven by cAMP in a positive feedback loop,
216
this establishes bistability. Clearly, using dual immunofluorescence staining,
217
simultaneous measurement of interacting protein expression and protein-protein
218
interaction (via reporter protein expression) can be successfully achieved at the
219
12 / 26
single-cell level using flow cytometry. It is worthy to note that although
220
pUT18C-Flag-pal and pKT25-His-tolB are two compatible plasmids, they have
221
distinct replication origins and different copy numbers (11). The plasmid of lower
222
copy number (pKT25-His-tolB) normally results in a lower level of protein expression,
223
which determines the rate of complex formation with its interacting protein partner.
224
Note that the same PMT voltage was used for the detection of His-tag and Flag-tag
225
signals, and the relatively higher signal for His tag is due to the higher affinity of
226
anti-his antibody.
227
228
Fig. 2. Simultaneous measurement of the expression of interacting proteins and 229
reporter protein by flow cytometry. (A) Bivariate dot-plot of green fluorescence 230
intensity (β-gal) versus red fluorescence intensity (His-tag) for E. coli BTH101 231
co-transformed with plasmids of A1) pUT18C/pKT25, A2) pUT18C-pal/pKT25-tolB, 232
and A3) pUT18C-Flag-pal/pKT25-His-tolB, respectively. (B) Bivariate dot-plot of 233
green fluorescence intensity (β-gal) versus red fluorescence intensity (Flag-tag) for E.
234
coli BTH101 co-transformed with plasmids of B1) pUT18C/pKT25, B2) 235
pUT18C-pal/pKT25-tolB, and B3) pUT18C-Flag-pal/pKT25-His-tolB, respectively.
236
Correlation between the Protein Interaction Strength and the Expression of
237
Interacting Proteins in a Single Cell. As illustrated in Figure 1, in the BACTH
238
system, expression of the β-gal reporter protein is regulated by the production of
239
cAMP and thus by the interaction of two hybrid proteins (10). We examined the
240
relationship between the β-gal reporter protein expression and His-TolB (the plasmid
241
with lower copy number in the cell) expression by flow cytometry. When the cultures
242
of E. coli BTH101 reached a sufficient cell density (OD ~1.5) after 12 h cultivation,
243
these cultures were co-transformed with plasmids of
244
pUT18C-Flag-pal/pKT25-His-tolB every two hours. Before immunofluorescent
245
staining, β-gal activity of each sample was assayed by a classical Miller’s protocol on
246
a spectrophotometer. Fig. 3A shows that β-gal activity increased from 375 at 12 h to
247
702 at 16 h and started to decrease slowly after then. Meanwhile, single cell
248
measurements by flow cytometry indicate that with the increase of cultivation time
249
from 12 h to 20 h, the fraction of cells expressing β-gal and His-TolB (Fig. 3B) kept
250
increasing from 7.5% to 75.5%. In contrast, the MFIs of positive cells decreased from
251
14500 to 2508 for β-gal signal (Fig. 3B1) and from 4155 to 495 for His-TolB signal
252
(Fig. 3B2), respectively. This phenomenon could be explained by the dilution of the
253
proteins inside a single cell upon cell division (19). When we plotted the MFI of β-gal
254
versus that of His-TolB after background signal subtraction for each protein, i.e.
255
(MFIβ-gal, P2-MFIβ-gal, P1) versus (MFIHis-TolB, P2-MFIHis-TolB, P1), a linear correlation with
256
R2 of 0.9995 was obtained (Fig. 3C), which suggests that in the BACTH system,
257
expression of the β-gal reporter protein is linearly proportional to the expression of
258
14 / 26
the interacting protein.
259
260
Fig. 3. Enzymatic and flow cytometric analysis of the reporter protein β-gal for E. coli 261
BTH101 co-transformed with plasmids pUT18C-Flag-pal/pKT25-His-tolB at different 262
cultivation time. A) Column chart of β-gal activity measured with a classical Miller’s 263
assay. B) Histograms of the fluorescence intensity distribution for β-gal and His-TolB 264
measured by flow cytometry. C) The correlation curve between the median 265
fluorescence intensities of β-gal and His-TolB after background subtraction.
266
The heterogeneity in the BACTH system has been well recognized because there
267
exist a big difference in plasmid copy number in different bacterial cells (11) and
268
stochasticity inherent in the biochemical process of gene expression (20). In order to
269
validate the generality of this observation, we first examined the correlation between
270
the expression of β-gal reporter protein and interacting protein for different colonies
271
at the same cultivation time. Fifteen different colonies were randomly picked from the
272
culture plate of E. coli BTH101 co-transformed with plasmids
273
pUT18C-Flag-pal/pKT25-His-tolB and inoculated in separate LB broth and cultivated
274
for 16 h. Fig. 4A shows the plot of β-gal expression versus that of interacting protein
275
His-TolB for different colonies, and a linear correlation was observed with R2 of
276
0.9789. These data along with those reported in Fig. 3 suggest that for an interacting
277
protein pair, there is an important heterogeneity in the expression of both the hybrid
278
and reporter proteins, but that the expression of β-gal reporter protein exhibits a linear
279
proportion to that of the hybrid proteins regardless of different cultivation time for a
280
single colony or different colonies at the same cultivation time. Clearly, these results
281
demonstrate that the expression of the β-gal reporter protein is not only affected by
282
the affinity of the interacting protein pair but also by the expression level of the
283
hybrid proteins. Therefore, we propose to use relative reporter protein expression
284
(RRPE), defined as the normalized β-gal expression to that of the interacting protein,
285
to estimate the interaction strength of protein pairs. As shown in Fig. 4B, the
286
measured RRPE (blue dots) for the Pal-TolB interaction pair exhibits a constant value,
287
whereas the β-gal activity (red dots) is much more diverging among different
288
colonies.
289
290
Fig. 4. Flow cytometric analysis of the expression of β-gal reporter protein and 291
interacting protein His-TolB for bacterial samples inoculated with different single 292
colonies. A) The correlation curve between the median value of β-gal fluorescence 293
intensity and that of His-TolB. B) Plot of the relative reporter protein expression 294
(RRPE) versus β-gal activity for 15 bacterial cultures inoculated with different single 295
colonies randomly picked from the plate co-transformed with plasmids 296
pUT18C-Flag-pal/pKT25-His-tolB.
297
16 / 26
Validation of the RRPE-BACTH Method for the Measurement of Protein
298
Interaction Strength. To investigate the potential of using RRPE-BACTH method
299
for evaluating the strength of protein-protein interactions, we compared the Pal-TolB
300
interaction by using two mutated forms of TolB along with the wild type TolB. These
301
two mutated TolB proteins bear truncations, one with four residues deleted from the
302
N-terminus (TolB Δ22-25) and the other with the entire N-terminal sequence deleted
303
(TolB Δ22-33), which lower the binding affinity to Pal (21). The dissociation constants
304
KD of these two truncated TolB proteins with Pal were reported to be 313 ± 15 nM
305
and 337 ± 18 nM, respectively, via ITC measurement at 30° C, which are about
306
tenfold higher than that of the wild type TolB (38 ± 3 nM). E. coli BTH101 cells were
307
co-transformed with pUT18C-Flag-pal/pKT25-His-tolB,
308
pUT18C-Flag-pal/pKT25-His-tolB Δ22-25, or pUT18C-Flag-pal/pKT25-His-tolB Δ22-33
309
plasmids and plated. Three individual colonies were picked and inoculated into LB
310
broth for each protein pair. These samples were immunofluorescence stained and
311
analyzed on the flow cytometer. Fig. 5A shows the representative bivariate dot-plots
312
of β-gal green fluorescence versus His-TolB red fluorescence and their fluorescence
313
distribution histograms for these three pairs. Fig. 5B indicates that the wild type TolB
314
exhibits the highest RRPE of 2.04 ± 0.21, and the mutated TolB Δ22-25 and TolB Δ22-33
315
shared comparable RRPE values of 0.88 ± 0.07 and 0.91 ± 0.06, respectively. Hence,
316
the RRPE value follows the change in binding affinity of the TolB/Pal interaction,
317
with larger RRPE corresponds to higher binding affinity in the BACTH system.
318
Therefore, RRPE may be used to assess the binding affinity of protein-protein
319
interaction in the BACTH system.
320
321
Fig. 5. RRPE measurement for Pal interacting with TolB and the two TolB mutants by 322
the BACTH system and flow cytometry. (A) The bivariate dot-plots of β-gal green 323
fluorescence versus His-TolB red fluorescence and the fluorescence distribution 324
histograms for wild type TolB and its two truncated forms: TolB Δ22-25 and TolB Δ22-33. 325
(B) Column chart of the RRPE for Pal interaction with TolB and the two truncated 326
forms. The error bar represents the standard deviation of three replicates.
327
To further validate the applicability of the RRPE-BACTH method in affinity
328
assessment, five pairs of acid (En) and base (Kn) α-helices with various heptad
329
repeats (n) that associate into coiled coils were constructed into the BACTH system
330
(Fig. 6A). E coil and K coil interacts through hydrophobic interaction at the interface
331
and electrostatic attraction between the oppositely charged residues from the helix,
332
and higher affinity is associated with longer helix (22). Among the five coiled-coils
333
chosen in the present study, the dissociation constants (KD) measured by surface
334
plasmon resonance (SPR) using a BIAcore were 30000 ± 3000, 7000 ± 800, 116 ± 8,
335
14 ± 1, and 0.063 ± 0.005 nM for the interactions of E3-K3, E5-K3, E4-K4, E5-K4,
336
18 / 26
and E5-K5, respectively (22). E. coli BTH101 cells were co-transformed with
337
pUT18C-Flag-En/pKT25-His-Kn plasmids and plated. For each protein pair, three
338
individual colonies were picked and inoculated into LB broth separately. These
339
samples were immunofluorescence stained and analyzed on the flow cytometer. Fig.
340
6B shows the representative bivariate dot-plots of β-gal green fluorescence versus
341
His-Kn red fluorescence along with the fluorescence distribution histograms for these
342
five interaction pairs. The measured RRPE values were 3.8 ± 0.1, 6.4 ± 0.5, 9.9 ± 0.1,
343
11.8 ± 0.3, and 5.9 ± 0.6 for E3-K3, E5-K3, E4-K4, E5-K4, and E5-K5, respectively.
344
It should be noted that taking advantage of high sensitivity of the RRPE-BATCH
345
method, we can discriminate interactions with KD lower than 104 nM. Fig. 6C shows
346
that the measured RRPE exhibited a strong correlation with the interaction affinity
347
from E3-K3 to E5-K4 except for E5-K5. This could be explained by the fact that
348
E5-K5 fits an interacting model with a relatively fast association and a very slow
349
dissociation, which is different from the other pairs (22). In contrast, in the BACTH
350
system, the synthesis of cAMP is irreversible and depends mainly on the rate at which
351
two coils associate (on-rate) to initiate complementation of T25 and T18. Because the
352
E5 coil was fused to T18 or T25 domain in the BATCH system, the on-rate of binding
353
with K5 could be reduced which leads to a lower binding affinity as demonstrated
354
with a decreased RRPE value (22). However, it needs to be pointed out that the
355
proposed RRPE method does not allow for the absolute quantification of protein
356
interaction affinity, because the BACTH system itself is an indirect method to detect
357
protein-protein interaction. Nonetheless, the good correlation between the measured
358
RRPE and the equilibrium dissociation constant reported in literature demonstrates
359
that the method is useful in providing a relative ranking of interaction strength for
360
protein variants in a given interacting pair of proteins.
361
362
Fig. 6. RRPE measurement for five pairs of coiled coil interactions using the BACTH 363
system and flow cytometry. (A) A schematic (adapted from the Fig. 1 by De 364
Crescenzo et al. (22)) of the acid (En)-base (Kn) coiled coils interaction with n 365
indicating the number of heptad repeats. (B) The bivariate dot-plots of β-gal green 366
fluorescence versus His tag red fluorescence intensity for E. coli BTH101 cells 367
co-transformed with plasmids of pUT18C-Flag-En/pKT25-His-Kn. (C) The correlation 368
of the RRPE values measured by flow cytometry for coiled coil interactions occurring 369
in the BACTH system with the affinity measured by SPR (22). The error bar 370
represents the standard deviation of three replicates.
371
Conclusion
372
20 / 26
We have developed a sensitive in vivo method for the quantitative measurement of
373
protein-protein interaction via the BACTH system and flow cytometry. Taking
374
advantage of the high-throughput and multiparameter measurement of single cell by
375
flow cytometry, the expression of reporter protein and interacting proteins can be
376
simultaneously measured and correlated at the single-cell level. It was found out that
377
for the BATCH system, for a bacterial culture inoculated even with a single colony
378
co-transformed with two plasmids encoding each of the two interacting proteins, there
379
exist two populations and a large heterogeneity for each population in protein
380
expression and reporter protein production, which would otherwise be masked by the
381
ensemble-averaged measurements. By measuring the expression level of interacting
382
protein and reporter protein for the population expressing β-gal, it was identified that
383
for an interacting protein pair, the value of RRPE is constant and is an intrinsic
384
feature. Moreover, a good correlation of the RRPE with the binding affinity of protein
385
pair was observed for Pal-TolB interaction with WT and mutant TolB proteins and for
386
several coiled-coil interactions. The RRPE method proposed here can not only be
387
used to validate existing protein interaction and finding new ones, but also to rank the
388
strength of interaction. It may further be used for highthroughput study of the binding
389
site of protein-protein complexes, selection of high-affinity antibodies, and screening
390
of peptide inhibitor libraries.
391
392
Materials and Methods
393
Reagents and Chemicals. Rabbit anti-β-galactosidase IgG was purchased from
394
Molecular Probes (Eugene, OR, USA). FITC-conjugated anti-His mouse monoclonal
395
antibody and goat anti-rabbit (GAR) IgG (H+L) were obtained from TransGen
396
Biotech (Beijing, China). DyLight-649-conjugated goat anti-mouse (GAM) IgG (H+L)
397
was purchased from EarthOx (San Francisco, CA, USA). Antibodies were diluted in 1%
398
fetal bovine serum (FBS) (obtained from Hyclone, Logan, Utah, USA) freshly
399
prepared in PBS before use. Enzymes used for molecular cloning were obtained from
400
TaKaRa Biotech (Dalian, China). Ortho-nitrophenyl-β-galactoside (ONPG),
401
lysozyme, GTE (50 mM Glucose, 25 mM Tris, 10 mM EDTA, pH 8.0), and X-gal
402
were purchased from Sangon Biotech (Shanghai, China). Paraformaldehyde (PFA)
403
stock solution (16%) was obtained from Alfa Aesar (Ward Hill, MA, USA). Other
404
reagents were purchased from Sinopharm Chemical Reagent (Shanghai, China). All
405
the buffers were filtered through a 0.22µm filter and used within three weeks.
406
Bacterial Strains and Plasmids. E. coli ER2738 was used for the cloning
407
experiments. The recombinant plasmids used in the present study are summarized in
408
Table S1 and were verified by sequencing. Oligonucleotides were synthesized by
409
Sangon Biotech and are listed in Table S2. Plasmid pKT25-His-tolB was constructed
410
by inserting the histidine tag (His-tag) into pEB362 at the PstI/EcoRI sites. The pal
411
gene was digested with EcoRI and XhoI from pEB356 and inserted into the
412
EcoRI/XhoI sites of pEB1030 to produce pUT18C-Flag-pal. Genes of tolB Δ22-25 and
413
tolB Δ22-33were amplified from plasmid pEB362 using suitable primers listed in Table
414
S2. The PCR products were cleaved by EcoRI and XhoI and cloned into the
415
EcoRI/XhoI sites of pKT25-His to yield plasmids pKT25-His-tolB Δ22-25 and
416
22 / 26
pKT25-His-tolB Δ22-33. To introduce Kn and En sequences into each BACTH vectors,
417
oligonucleotide primers of Kn and En that are complementary to each other were
418
synthesized as listed in Table S2 and annealed by heating at 95 °C for 5 min.
419
Followed by cooling to room temperature, the products with cohesive ends were
420
inserted into plasmids pKT25-His and pUT18C-Flag at the EcoRI/XhoI sites to obtain
421
plasmids pKT25-His-Kn and pUT18C-Flag-En, respectively.
422
E. coli BTH101 (F-, cya-99, araD139, galE15, galK16, rpsL1, hsdR2, mcrA1,
423
mcrB1) was used as the reporter strain of the BACTH system. Competent E. coli
424
BTH101 strains co-transformed with two-hybrid plasmids bearing two different
425
antibiotic resistances were spread on Luria-Bertani (LB) plates containing 100 µg/mL
426
ampicillin, 50 µg/mL kanamycin, and 40 µg/mL X-gal at 30 °C. After incubating the
427
plates for 2 days, single colonies with successful co-transformation of two hybrid
428
plasmids were picked and inoculated in 2 mL of LB containing 100 µg/mL ampicillin
429
and 50 µg/mL kanamycin. Cultures were grown overnight with shaking (250 rpm) at
430
30 °C, unless specified otherwise. The harvested bacterial sample was adjusted to
431
OD600 ~1.0, immunofluorescently stained and analyzed on the flow cytometer.
432
Immunofluorescent Staining. To a 200 µL of the harvested bacterial cells, 8 µL of 1
433
M NaPO4 (pH 7.4) and 40 µL of the primary fixative buffer (3 µL of 25%
434
gluteraldehyde per mL of 16% paraformaldehyde) were added and incubated at room
435
temperature for 15 min followed by 30 min on ice. The sample was washed twice
436
with 200 µL PBS and resuspended in 50 µL PBS. Then, 500 µL of ice cold 80%
437
methanol was added and the sample was treated for 1 h at room temperature. The
438
sample was washed twice with GTE buffer. The cells were permeabilized by
439
resuspending in 100 µL of 2 mg/mL lysozyme in GTE and incubated for 10 min at
440
room temperature. After washing twice with PBS, the cells were blocked in 100 µL 1%
441
FBS for 10 min. Then 20 µL of the suspension was centrifuged and resuspended in 40
442
µL of 5 µg/mL rabbit anti-β-gal antibody with/without 5 µg/mL mouse anti-His/Flag
443
antibody depending on the experimental requirement. After 1 h incubation at room
444
temperature, the sample was centrifuged and washed with PBS, then resuspended in
445
40 µL of 10 µg/mL FITC-conjugated GAR antibody with/without
446
DyLight-649-conjugated GAM antibody. The suspension was incubated for 30 min at
447
room temperature, centrifuged, and resuspended in 50 µL PBS. For flow cytometry
448
analysis, the sample should be diluted 500-fold with PBS before loading.
449
Flow Cytometric Measurement. A Becton Dickinson FACSVerse flow cytometer
450
equipped with 488 nm and 640 nm excitation lasers was used in this study. FL1
451
(527/32 nm band-pass filter) channel and FL2 (660/10 nm band-pass filter) channel
452
were used to detect the fluorescence of FITC and DyLight 649, respectively for the
453
immunofluorescently stained bacteria. A threshold value of 200 was set on FL1 to
454
eliminate non-bacterial particles. A total of 10000 events falling in the gated region
455
were collected for each sample. Data acquisition and analysis were carried out by
456
using BD FACSuite software. The data were analyzed by Flowjo 7.6.1 software (Tree
457
Star, Inc., Ashland, OR).
458
Measurement of β-Gal Enzyme Activity by ONPG Colorimetric Assay. A
459
protocol described in the literature was followed (12). Briefly, 200 µL of the
460
24 / 26
harvested bacterial cells (OD600 ~1.0) were treated by 3 µL toluene and 3 µL 0.01%
461
SDS (shaking at 37 °C for 30 min). Then 1.8 mL PM2 buffer (70 mM Na2HPO4·12
462
H2O, 30 mM NaH2PO4·H2O, 1 mM MgSO4, 0.2 mM MnSO4, pH7.0) with 100 mM
463
β-mercaptoethanol was added and mixed thoroughly. After that, 250 µL of the ONPG
464
substrate solution (4 mg/mL ONPG in PM2 buffer without β-mercaptoethanol) was
465
added to 1 mL of the mixture. The enzymatic reaction was carried out immediately on
466
a DU-800 spectrophotometer (Beckman Coulter) with measurement of OD420 nm. The
467
β-galactosidase activity corresponds to 200 × (OD420 nm, t2 - OD420 nm, t1) / (t2-t1) (min)
468
× 10. The factor 200 is the inverse of the absorption coefficient of o-nitrophenol,
469
while the factor 10 is the dilution factor.
470 471
ACKNOWLEDGEMENTS
472
We acknowledge support from the National Natural Science Foundation of China
473
(21105082, 21225523, 91313302, 21027010, 21475112, 21472158, and 21521004),
474
and the Program for Changjiang Scholars and Innovative Research Team in
475
University (IRT13036), for which we are most grateful.
476
Author contributions
477
L.W., X.W., and X.Y. conceived and designed the research; L.W., X.W., T.L.,
478
and J.Z. performed research; L.W., X.W., and X.Y. analyzed data and wrote the
479
paper.
480
Conflict of interest statement
481
The authors declare no conflict of interest.
482